Introduction
Our knowledge of how cancer cells respond and subsequently develop resistance to chemotherapy is far from complete. Two recent studies in Nature1, 2 and one in Cancer Research3 unravel how such resistance develops in ovarian cancers associated with mutations in the BRCA1 and BRCA2 genes. These mechanistic findings may have implications for how chemotherapeutic agents are prescribed for cancer patients and demonstrate the need for understanding whether targets of chemotherapy are necessary for cancer cell maintenance.
The FANC-BRCA pathway consists of a number of proteins that are required for an appropriate cellular response to DNA cross-linking, a form of DNA damage. An inherited mutation inactivating either the BRCA1 gene or the BRCA2 gene can be found in people with familial breast and ovarian cancer. In the tumor tissue itself, the remaining normal copy of the BRCA-encoding gene is typically lost. BRCA1 and BRCA2 are therefore classified as 'tumor suppressor genes', because their loss results in the development of cancer. The BRCA1 and BRCA2 genes encode proteins that are involved in homologous recombination, a process that repairs double-stranded DNA breaks and stalled DNA replication forks. As a result of BRCA deficiency, mutations that lead to the development of cancer can accumulate. The lifetime risk of developing ovarian cancer approaches 40% for women with BRCA1 mutations and is 10–20% for women with BRCA2 mutations4.
Owing to a lack of both symptoms and effective screening tools, most individuals with ovarian cancer are diagnosed with advanced-stage disease, which requires chemotherapy for disease control. Encouragingly, cells that are deficient in repairing double-stranded DNA breaks are particularly sensitive to chemotherapeutic agents that work by inducing such breaks, such as cisplatin, presumably because accumulating mutations eventually have deleterious consequences leading to cell death. A new class of drugs undergoing early clinical development inhibits poly(ADP-ribose) polymerase (PARP), a protein required for repairing single-stranded breaks. Without PARP, these single-stranded breaks also stall DNA replication forks; PARP inhibitors therefore selectively kill BRCA-deficient cells5, 6 (Fig. 1) and may be a less toxic form of effective therapy.
Figure 1: BRCA2 as a therapeutic target.
(a) Loss of BRCA2 results in ineffective DNA repair, which in turn results in the generation and accumulation of mutations that cause cancer. (b) BRCA2 deficiency can be exploited therapeutically by agents that lead to double-stranded DNA breaks, such as cisplatin. Substantial DNA damage results in the selective cell death of BRCA-deficient cells. (c) PARP can be targeted with small molecule inhibitors, leading to an accumulation of single-stranded DNA breaks, which can subsequently result in lethal double-stranded DNA breaks in the absence of BRCA2. Not shown: Restoration of BRCA2 renders cells insensitive to cisplatin and PARP inhibitors.
Full size image (22 KB)The long-term success of drugs such as cisplatin is severely limited by incomplete disease eradication (resulting in the need to repeatedly retreat patients who relapse) and, more importantly, by the eventual development of drug-resistant disease; as a result, most advanced-stage patients are not cured.
The three new studies examine how resistance to cisplatin or PARP inhibitors develops in primary ovarian cancers and pancreatic cancer cell lines associated with mutations in BRCA1 or BRCA2. Remarkably, resistance correlated with restoration of detectable levels of BRCA1 and BRCA2 protein as a result of secondary mutations that restore the reading frame of the proteins. The new findings dovetail with previous work by D'Andrea and his colleagues7, who found that cisplatin hypersensitivity in some sporadic ovarian cancer cell lines is associated with lack of an intact FANC-BRCA pathway, whereas cell lines that developed cisplatin resistance were associated with restoration of this pathway through demethylation of the FANCF gene7.
Because cisplatin is mutagenic, it is plausible that the observed mechanism of resistance to cisplatin is partly induced by the therapy itself. Alternatively, the same BRCA deficiency that prompted genomic instability and the initial cancer development may also lead to BRCA mutation correction. These studies provide clinical evidence for the model of D'Andrea and his colleagues7, hypothesizing that impaired BRCA1 or BRCA2 function is not required for tumor maintenance, a sobering finding given the excitement surrounding the development PARP inhibitors.
The new studies reveal the capability of human malignancy to initially shut off cellular machinery to facilitate initial cancer establishment and then to creatively restore the machinery and thereby generate resistance to therapy. The term 'tumor suppressor gene' is therefore misleading in this context, as it implies that restoring the function of BRCA proteins may lead to suppression or death of cancer cells, and for BRCA1 and BRCA2 this is clearly not the case. From a therapeutic target perspective, it is therefore more useful to conceptually subdivide this group of genes into those that are involved in both initiation and maintenance of cancer cells when lost and those that are essential for cancer initiation alone.
There are clear similarities between BRCA1 and BRCA2 and BCR-ABL, the hallmark of chronic myeloid leukemia (CML) that has been successfully targeted with a selective inhibitor, imatinib8. Both types of mutations are early genetic events and confer sensitivity to therapy, secondary mutations in these genes result in acquired resistance to therapy9 and both types of mutations result in increased genomic instability10.
But there are also some key differences to bear in mind. Whereas functional loss of BRCA1 or BRCA2 can lead to an accumulation of genetic events that result in cancer, once the disease is established, BRCA deficiency is no longer required or advantageous. In contrast, oncogenic 'driver' mutations such as BCR-ABL are important not only for the initiation of disease, but also for its maintenance—apparently even in the late phases of CML, which are associated with an accumulation of cooperating oncogenic mutations yet can respond deeply (although typically transiently) to kinase inhibitor therapy. Identification of molecular pathways that drive BRCA1 or BRCA2 deficiency–associated cancers will probably be necessary to substantially increase the rate of deep and durable clinical responses, and these three studies provide a compelling rationale for testing combinations of cisplatin with agents that target pathways that are essential for cancer cell maintenance. Some studies suggest that targeting kinases such as human epidermal growth factor receptor-2 or vascular endothelial growth factor may be fruitful in this population4.
The propensity of BRCA deficiency and BCR-ABL to facilitate genomic instability, and thus therapy-resistant disease, can hopefully be minimized with the use of effective therapy. Indeed, it seems that the prognosis of patients with CML on imatinib improves with increasing duration of therapy (patients are less likely to suffer relapse as time goes on)11. This observation suggests that persistent BCR-ABL inhibition is effectively minimizing any contribution that BCR-ABL–mediated genomic instability can confer. The correction of a BRCA1 or BRCA2 defect may similarly limit the ability of resistant cells to develop resistance to subsequent therapies. However, if cisplatin is withdrawn at the time of resistant disease, it is plausible that BRCA-deficient cells will again become detectable, which would be expected to generate further genetic complexity in the tumor.
There may thus be a therapeutic rationale for continuing cisplatin therapy (in addition to second-line chemotherapeutics) despite the presence of cisplatin-resistant disease, in an effort to continue negative selective pressure on BRCA-deficient cells, although clinical trials will be necessary to formally address this issue. Similarly, one can envision that some BRCA-restored cisplatin-resistant cells that become resistant to subsequently administered chemotherapeutic agents may also have again lost BRCA and may therefore respond again to cisplatin. A similar phenomenon has been observed in patients with CML treated with kinase inhibitors12.
Importantly, it may be easy to determine the BRCA status of cells by analyzing the cellular localization of RAD51, a BRCA-associated protein, after generation of a double-stranded DNA break and to guide further therapeutic decisions accordingly.
Given that patients frequently relapse with cisplatin-sensitive disease before eventually developing cisplatin resistance, an important issue involves the optimal duration of chemotherapy. Currently, it is accepted clinical practice to administer four to six cycles of chemotherapy and halt treatment once patients enter a clinical remission. Although there may be some benefit to 'maintenance' chemotherapy with select chemotherapeutics (for example, paclitaxel), a small number of studies have been unable to show a survival benefit of prolonged cisplatin therapy, perhaps in part owing to the design of the studies13.
The new findings by Ashworth, Taniguchi and their colleagues1, 2, 3should rekindle a more rigorous evaluation of maintenance therapy with agents that have activity against BRCA-deficient cells such as cisplatin or, possibly, PARP inhibitors. It is hoped that such a strategy will substantially reduce or potentially eliminate minimal residual disease and improve therapeutic outcomes.
